Migration, distribution, and diving behavior of adult male loggerhead sea turtles (Caretta caretta) following dispersal from a major breeding aggregation in the Western North Atlantic
- First Online:
- Cite this article as:
- Arendt, M.D., Segars, A.L., Byrd, J.I. et al. Mar Biol (2012) 159: 113. doi:10.1007/s00227-011-1826-0
- 439 Views
Sixteen satellite-tagged adult male loggerhead sea turtles (Caretta caretta) dispersed widely from an aggregation near Port Canaveral, Florida, USA (28°23′N, −80°32′W) after breeding. Northbound males migrated further (990 ± 303 km) than southbound males (577 ± 168 km) and transited more rapidly (median initial dive duration = 6 (IQR = 4–16) versus 19 (IQR = 10–31) min, respectively).. Migration occurred along a depth corridor (20–40 m) except where constricted by a narrow continental shelf width. Males foraged in areas 27 ± 41 km2 day−1 at locations <1–80 km from shore for 100.1 ± 60.6 days, with variability in foraging patterns not explained by turtle size or geography. Post-breeding dispersal patterns were similar to patterns reported for adult female loggerhead sea turtles in this region and adult male loggerhead sea turtles elsewhere in the northern hemisphere; however, foraging ground distributions were most similar to adult female loggerhead sea turtles in this region.
Long-distance migrations by adult males are predominantly reported from oceanic habitats (Plotkin et al. 1996; James et al. 2005; Plotkin 2010) and insular regions (van Dam et al. 2008; Schofield et al. 2010a). However, despite their importance and decades of sea turtle research, little is known about seasonal movements of adult males. Seasonal distribution data for adult male sea turtles are limited to a few global study areas for each of five species, with observations collected from relatively small sample sizes in all studies (Plotkin et al. 1996; Plotkin 2010; Hatase et al. 2002a; James et al. 2005; Shaver et al. 2005; van Dam et al. 2008; Hays et al. 2010; Schofield et al. 2010a, b). In contrast, the movements of post-nesting females, notably their use of bathymetric and hydrographic cues (reviewed by Luschi et al. 2003) and transit through migratory corridors (Morreale et al. 1996; Seminoff et al. 2008; Hawkes et al. 2011), have been extensively documented by satellite telemetry.
Foraging habitat dichotomy is suggested by satellite telemetry for adult loggerhead sea turtles in the northern hemisphere. At the largest rookery in the world on the shores of the Arabian Sea, larger adult females forage in continental shelf seas after nesting while smaller adults forage off the shelf in oceanic habitats (Rees et al. 2010). Similar patterns are also reported for adult females from tropical latitudes in the Eastern Atlantic Ocean (Hawkes et al. 2006) and from temperate latitudes in the Western Pacific Ocean (Hatase et al. 2002b, 2010). Size-based foraging dichotomy is also reported, although to a lesser extent, for adult male loggerhead sea turtles in the Western Pacific Ocean (Sakamoto et al. 1997; Hatase et al. 2002a). Satellite telemetry has also revealed inshore and pelagic foraging distributions for adult male loggerhead sea turtles in the Ionian and Adriatic Seas, but evidence of size-dependent behavioral polymorphism was not suggested (Schofield et al. 2010a).
Recent findings of adult loggerhead sea turtle foraging dichotomy in the northern hemisphere come from studies conducted at locations where continental shelves are narrow and as such, may not be generalized to sea turtles foraging over wide continental shelves. For example, use of oceanic habitats off the continental shelf appears to be the exception rather than the rule for adult female loggerhead sea turtles in the North Western (NW) Atlantic Ocean (Keinath 1993; Plotkin and Spotila 2002; Hawkes et al. 2011). Much of what is known about male distributions in this region stems from rare sightings or isolated collections at locations ranging from lagoons to across nearly the entire continental shelf. Year-round occurrence has been reported for southerly (<25°N) latitudes from sightings and satellite telemetry (Hickerson and Peccini 2000; Schroeder et al. 2003), with seasonal occurrence on the inner continental shelf reported from more northerly (32–37°N) latitudes (Van Dolah and Maier 1993; Mansfield 2006). Because of greater near shore sampling effort, male residency patterns further from shore may be under-reported.
Satellite telemetry was used to monitor the post-breeding movements of reproductively active (Blanvillain et al. 2008) adult male loggerhead sea turtles collected from an annual breeding aggregation (Henwood 1987; Wibbels et al. 1987). The first objective was to test the hypothesis of non-random migration away from a breeding ground located 40 km north of the second largest loggerhead rookery in the world (NMFS and USFWS 2008). The second objective was to test the hypothesis of size-related foraging dichotomy at post-breeding foraging grounds. Because of the greatly expanded continental shelf associated with subsiding continental margins that characterize the south and east coasts of North America, we predicted a lack of size-related foraging dichotomy as seen in previous studies. Lastly, we wanted to test whether diving behavior was consistent across latitudinal gradients during transit (objective three) and among foraging grounds (objective four) in order to determine the extent to which generalizations could be made regarding adult male loggerhead sea turtle diving behavior after the breeding period.
Materials and methods
Sea turtle collection and tagging
Adult male loggerheads were captured by trawling in the shipping entrance channel at Port Canaveral, Florida, USA (28°23′N, −80°32′W) in April 2006 and 2007. Minimum straight-line carapace length (SCLmin, cm) and body mass (kg) were recorded, and sex was confirmed via laparoscopy (Blanvillain et al. 2008). Satellite transmitters (ST-20, Model A2020; Telonics, Inc., Mesa, Arizona) were attached as described by Arendt et al. (in press). Males were released 3.8 ± 1.9 km (mean ± standard deviation [SD]) from where they were captured, with seven males released 7.9 ± 2.2 h after initial collection while nine males were held overnight and released 21.8 ± 2.0 h after initial collection. Satellite telemetry data were collected until the satellite tags stopped transmitting.
Data collection and processing
The duty cycle for transmitters was set to continuous on with a 45-s repetition rate at 401.630–401.656 MHz; however, a saltwater switch (with 4 h failsafe duration) conserved battery life while turtles were submerged. In addition to location, ambient temperature at each surfacing and three dive cycle metrics corresponding to the number of dives, mean dive duration (s), and the percent of time submerged during 4, 6-h data bins per day (1900–0059 h EST; 0100–0659 h; 0700–1259 h; 1300–1859 h) were recorded, with archived data for each prior bin transmitted during each ‘active’ data bin.
Telemetry data were processed and distributed by CLS America (Largo, Maryland) and were archived using the Satellite Tracking and Analysis Tool (STAT; Coyne and Godley 2005), and the latter paired turtle locations with bathymetry data (m) and calculated distance (km) to the nearest shoreline. Locations with elevations >1 m above sea level (filter criteria used by Mansfield et al. 2009 and Schofield et al. 2010a) were removed which represented 8.3% of the original data set. Remaining locations were subjected to inner turning angle (5°) and speed filters created by the authors (Arendt et al. in press, Supplement 1). The speed filter was set to 3 km h−1 on foraging grounds, but liberally set to 15 km h−1 in transit given documentation of speeds >10 km h−1 for this species (Renaud and Carpenter 1994; Papi et al. 1997).
During transit, all transmitted locations were subjected to filtering. Given serial correlation of foraging ground locations and the need to standardize data for spatial analyses (Swihart and Slade 1985; Makowski et al. 2006), on the foraging ground, filters were applied to a single location per turtle per day. Similar to methods used by Hawkes et al. (2011), an initial daily ‘best’ location for each turtle was selected as the first occurrence of the most accurate location class according to the CLS/Argos hierarchy (3 > 2 > 1 > 0 > A > B > Z). Spurious locations identified by these filters were removed, and the ‘next best’ location systematically substituted until filter criteria were met. If substitute locations were not available or could not pass filter criteria, no location data for that turtle for that day (i.e., a turtle day) were retained. During transit, 80 ± 8% (mean ± SD) of individual turtle locations were retained, with 69% of deleted locations removed due to the inner turning angle filter and elevation and speed filters accounting for nearly equal (15–16%) rejection of all other locations. On foraging grounds, location data for 92 ± 9% of original observation days per turtle were retained.
Temperature records >50 and <5°C were removed (n = 215 observations, 0.8% of the data set) given high improbability of occurrence during the study. Data for the number of dives, mean dive duration, and submergence time were removed when mean dive duration (s) multiplied by the number of dives exceeded the possible seconds in 6 h (n = 4,186 observations, 19% of the data set). When replicate observations for a dive metric occurred during a data bin, the mode was selected to provide a single observation per unique bin. When no mode was observed, the mean was computed if the maximum difference among replicate values was <5% relative (dive duration) or <5% absolute (surface interval), otherwise all replicate observations were removed. Total bin retention (mean ± SD) for dive duration and surface interval data was 80 ± 6% and 81 ± 5%, respectively.
Location data during (a) transit and (b) residence at a distant foraging ground were analyzed separately for northbound and southbound adult male loggerhead sea turtles, as only resident males migrated east (Arendt et al. in press). Transit was defined as the period between departure and subsidence of directed latitudinal movement as documented in other studies (James et al. 2005; Hawkes et al. 2011) and termed type A1 movements by Godley et al. (2008). Turtles were considered resident at a foraging ground immediately after the transit period until the turtle was no longer detected at this location or transited to a subsequent location.
Statistical analyses (α = 0.05) were performed using Minitab 15® (Minitab, Inc., State College, Pennsylvania) except for a Fisher’s exact test to compare migration direction (2006 vs. 2007) performed in Vassar Stats (http://faculty.vassar.edu/lowry/fisher.html). Gross transit distances (km) and rates (km day−1) between the Port Canaveral shipping channel and distant foraging grounds were compared using two-sample t tests. Gross transit distance was calculated as a straight-line distance between the Port Canaveral shipping channel and the median (to estimate central tendency) latitude and longitude where adult male loggerhead sea turtles occurred at distant foraging grounds. Gross transit rates were calculated as this distance divided by the number of days in transit. Ambient temperature distributions of north- and southbound males during transit were statistically compared with a Chi-square contingency test.
The distance from shore (km) of the 100-fathom contour was computed at 0.25° increments in order to quantitatively evaluate migration routes relative to the width of the continental shelf. Straight-line distance to shore was measured (mm) across latitude lines on nautical charts (#11009, #11006, #11013, and #12200), converted to longitude based on chart scale, and then converted to distance (km) by multiplying by 96.49 (km deg−1 of longitude at 30°N). The maximum range in male longitude distributions at the same 0.25° latitude increments was expressed as a proportion of the continental shelf occupied and analyzed with linear regression.
Water depth (m), distance from shore (km) and rate of travel (km h−1) between filtered locations were compared among north- and southbound males using Kruskal–Wallis (K–W) analysis of variance by ranks and Bonn–Ferroni (B–F) pairwise comparisons given non-normal distribution of data and their residuals. Given constriction of continental shelf width between 34.7 and 36.0°N, northbound male transit data were analyzed separately south of, within, and after this shelf width constriction zone. Similarly, southbound male transits were partitioned and analyzed spatially to assess potential influences of continental shelf constriction (and proximity to the Florida current) during transit south to 26°N, within the Florida Straits (24.8–26°N), and in the Gulf of Mexico (GOM) north of 24.8°N.
Minimum Convex Polygons (MCP) on the foraging grounds were calculated (km2) around all daily locations for each male using the MCP tool in the Spatial Ecology toolset in the software package Geospatial Modelling Environment (http://www.spatialecology.com/gme/index.html). Because the number of daily locations used to calculate an MCP varied substantially among turtles, the MCP area was divided by the number of locations used to calculate an MCP in order to standardize comparisons to a relative scale (i.e., km day−1). Cluster analysis (Euclidean distance single linkage) was used to identify associations between standardized foraging area and SCLmin (cm), body mass (kg), median distance from shore (km), median water depth (m), median latitude (degrees), median travel rate on the foraging ground (km h−1), and the number of locations used to calculate the MCP. Linear regression was used to statistically evaluate all variable clusters with foraging area and between carapace length and distance from shore.
Neither dive duration nor surface interval data (nor their residuals) were normally distributed; thus, K–W analyses and B–F comparisons were used to statistically compare these variables between north- and southbound males during transit and on foraging grounds. Cluster analysis was also used to identify variable associations with dive duration and surface intervals during transit (north- and southbound males analyzed separately) and on foraging grounds (pooled data). Seven variables were evaluated for clustering with these two dive cycle metrics: body mass (kg); time of day (data bin); water depth (m); distance from shore (km); rate of travel (km h−1); ambient temperature (°C); and either Julian day (foraging ground) or the percent of transit expired calculated as the sequential transit day divided by the total number of days in transit. Daily means for water depth, distance from shore and rate of travel (and SD for travel rate) were computed from multiple filtered locations recorded on each transit day, then repeated for up to four bins of dive metric data per day. However, because a single water depth, distance from shore and rate of travel was associated with a single (daily) filtered location on foraging grounds, these values were repeated for up to four dive cycle metric data bins per day. Daily mean temperature was calculated from multiple observations for each turtle day during transit and on foraging grounds, with temperature SD also included in the cluster analysis for transit. K–W tests and B–F comparisons were used to statistically evaluate all clusters that occurred with dive duration or surface interval data. With the exception of time of day, which was already partitioned into four bins, all other variables were grouped ad hoc with an emphasis on even appropriation of samples prior to performing statistical tests.
Demographic (SCLmin, cm; body mass, kg) and distribution data for 16 migrant adult male loggerhead sea turtles satellite-tagged after collection from the Port Canaveral, Florida shipping entrance channel in April 2006 and 2007
n Gulf of Mexico
n Gulf of Mexico
n Gulf of Mexico
Travel routes were constructed from 1,489 filtered locations, of which 19% were comprised of location classes 1–3, 31% were from location classes 0 and A, 48% were location class B, and less than 2% of locations were location class Z (Supplement 1). Northbound males migrated along a continental shelf that was 131 ± 42 km wide, and <80 km wide only at the onset of transit (<29°N) and between 35 and 36°N. Northbound migratory routes were spread across 70 ± 38% (mean ± SD) of the width of the continental shelf, but overlapped extensively at latitudes associated with a constricted continental shelf (Fig. 1). Longitudinal distribution decreased significantly at higher latitudes (F1,41 = 18.7, r2 = 0.30, P < 0.001), but was also significantly auto-correlated (F1,41 = 135.0, r2 = 0.76, P < 0.001) with fewer males at the highest latitudes. In contrast, southbound males migrated along a continental shelf that ranged from 55 km wide at the onset of transit to <10 km wide through the Florida Keys (24.5°N), but increased to 204 ± 75 km wide in the GOM where three males were detected on foraging grounds. Southbound route overlap was only noted in the Florida Keys where the continental shelf was narrow (Fig. 1).
Adult male loggerhead sea turtles remained at primary foraging grounds for 26 to 248 days until detection ceased (n = 12) or until they relocated (n = 3; Table 1). Data collection was significantly longer for five southbound males (T13 = −2.49, P = 0.027; n = 502 locations) distributed between latitudes 23.2 and 30.3°N than for 10 males (n = 661 locations) that foraged between latitudes 32.3 and 39.4°N. Northbound males were detected on primary foraging grounds for 77 ± 36 days (mean ± SD), and southbound males were monitored on primary foraging grounds for 147 ± 76 days. Adult male loggerhead sea turtles remained at primary foraging grounds at mean temperatures of 16.8–30.7°C (Fig. 2) that were auto-correlated with latitude (F1,13 = 86.3, r2 = 0.85, P < 0.001).
Thirty-three percent of location data analyzed on primary foraging grounds was associated with location classes 1–3, 34% with location classes 0 and A, 33% with location class B, and <1% with location class Z (Supplement 1). Primary foraging ground (MCPs) spanned 31 km2 (male 64,543, n = 19 locations) to 3,234 km2 (male 64,548, n = 62 locations); however, when standardized by the number of input locations (n = 6–180; mean ± SD = 78 ± 53) forage areas ranged from 1.6 to 166.8 km2 day−1. Northbound males were associated with larger (42 ± 53 km2 day−1) areas than southbound males (15 ± 11 km2 day−1), but differences were not statistically significant (T13 = 1.12, P = 0.283).
The strongest (76% similarity) cluster analysis relationship on foraging grounds was between travel rate and distance from shore, which was significant (F1,13 = 5.0, r2 = 0.22, P = 0.044). Cluster analysis also revealed moderate associations between foraging area and latitude (73% similarity) and with SCLmin (62% similarity); however, regressions were not significant (P = 0.091 and P = 0.921, respectively). Median distance from shore was moderately (72% similarity) clustered with SCLmin; however, this regression was also not significant (P = 0.278).
Only one male was detected at a secondary foraging ground during summer (Figs. 1, 2). Between 31 August and 6 September 2007, male 64,548 traveled 420 km from the middle continental shelf off New Jersey (39.3°N) to the inner shelf near Cape Hatteras, NC (35.3°N). Through 30 September, this male was detected (n = 17 locations) within an area of 214 km2 (13 km2 day−1) which was 80% smaller than the area it occupied off New Jersey, but comparable to males 73100 (20 km2 day−1; n = 61 locations) and 73098 (8 km2 day−1; n = 98 locations) that were detected at the same general location until 2 August and 17 September, respectively.
Location data were collected for three adult males detected during winter. Male 73101 remained on the same foraging ground in the nGOM (30.3°N) from 22 June 2007 until detections ceased on February 22, 2008. Two other males migrated 262 (male 73111) to 364 km (male 73096) to the south of seasonal foraging grounds to over-winter, partially or completely, in areas twice as large as associated with their respective foraging grounds (Table 1; Fig. 1). Ambient temperature distribution (mean ± SD) for all three males monitored between 21 December and 21 March was 18.4 ± 2.4°C (n = 530 observations).
Diving and surfacing behavior
Time spent at the surface was statistically different (H5 = 69.4, P < 0.001) among adult male loggerhead sea turtles that migrated north (median = 4.6%, IQR = 3.3–5.9%, n = 537 bins) compared to males that migrated south (median = 4.1%, IQR = 3.0–5.3%, n = 296 bins). The greatest proportion of time spent at the surface during transit occurred north of 36°N (median = 6.0%, IQR = 4.2–9.0%, n = 57 bins), but was statistically similar to the amount of time spent at the surface by males in transit in the GOM (Fig. 4).
Cluster analysis (n = 655 bins) revealed the strongest association between surface interval and dive duration for northbound males (59% similarity). Significantly (H1 = 14.17, P < 0.001) less time was spent at the surface by northbound males that dove <5 min (median = 3.5%, IQR = 2.2–4.7%, n = 215 observations) than northbound males that dove >5 min (median = 5.4%, IQR = 4.4–6.4%, n = 264 observations). For southbound males, the strongest surface interval association (67% similarity) was with transit location.
Species propagation is contingent upon successfully locating and copulating with mates, which effectively leaves reproductively mature individuals of the opposite sex two options: remain in proximity throughout the year or reconvene at spatially and temporally established locations. With respect to loggerhead sea turtles in the NW Atlantic Ocean, the data contributed by this study, when examined in the context of patterns reported for adult females (Plotkin and Spotila 2002; Hawkes et al. 2011, suggest the use of both strategies. After breeding (Blanvillain et al. 2008) in a historically (Henwood 1987; Wibbels et al. 1987) large aggregation, adult males dispersed to foraging grounds spanning more than half of this specie’s temperate latitudinal range in the NW Atlantic (Watson et al. 2005). Two-thirds of migrant adult males foraged in areas associated with seasonal foraging grounds for females nesting north of Florida, with substantial overlap north of 35°N (Plotkin and Spotila 2002; Hawkes et al. 2011). Three of five males that migrated south foraged on either side of the Florida Straits in the same general areas preferred by post-nesting females from multiple nesting assemblages in Florida (Foley et al. 2008; Girard et al. 2009). Foraging by adult loggerhead sea turtles in the northern GOM has only been reported for two males (present study) and two females (Girard et al. 2009) to date.
Temporal overlap in spatial distribution of adult males and females is likely non-continuous on an annual basis due to asynchronous peaks in reproductive activity and subsequent post-breeding dispersal patterns. Migrant adult male loggerhead sea turtles dispersed away from Port Canaveral in May, concurrent with post-breeding dispersal of resident males (Arendt et al. in press). All migrant males arrived at post-breeding foraging grounds in June, the apex of nesting activity for loggerhead sea turtles in the NW Atlantic (NMFS and USFWS 2008). Given that the breeding aggregation from which these males were collected was located 40 km north of the second largest loggerhead sea turtle rookery in the world (NMFS and USFWS 2008), it is unlikely that females that bred with males in the present study also migrated north with said males. Alternatively, given significantly longer dive durations of southbound male migrants, it is possible that they initially followed females to the rookery before fully initiating their migration. However, on a larger scale, given the diversity of male migration routes observed, mixed sex migration of adult loggerhead sea turtles from breeding to foraging grounds was not suspected. Nevertheless, subsequent co-habitation at predictable foraging grounds (Hawkes et al. 2011) after completion of breeding activities by each sex increases the probability of males relocating females after swimming hundreds of kilometers to distant foraging grounds, despite a seasonal lag in arrival times. As such, we suggest that although breeding occurs in spring, on an annual basis, adult male and female sea turtles likely spend significant time in general proximity.
Spatial corridors provide an additional mechanism for adult males to relocate adult females after bouts of separation. Similar to breeding grounds, corridors therefore represent critical conservation zones. Migratory males funneled through two such corridors east of ‘island chains’ inshore of narrow continental shelf margins between 34.7 and 35.3°N and between 24.5 and 25.9°N. Both corridors are used by females to reach post-nesting foraging grounds (Plotkin and Spotila 2002; Foley et al. 2008; Hawkes et al. 2011). The northern corridor is also re-used by females to reach over-wintering destinations (Keinath 1993; Mansfield 2006; Hawkes et al. 2011). Two males that foraged north of this corridor transited south through it between September and November, consistent with female re-migration southward in October (Hawkes et al. 2011). Contact with all other males that foraged north of this corridor was lost in August; however, they likely over-wintered to the south given unsuitable conditions north of 35°N (Hawkes et al. 2011).
The diversity of loggerhead sea turtle distribution with respect to distance from shore and water depth suggested a high degree of suitable foraging habitats; however, predictive indicators were not identified to account for differences in geographic habitat selection by individual males. One interpretation of this finding is that the magnitude of differences in water depth, distance from shore, and foraging area size were too subtle to be detected. For example, size-related dichotomy may only be detectable between differences on the magnitude of oceanic versus neritic foraging, which has thus far only been reported from volcanic island chains (Hatase et al. 2002a, b; Hawkes et al. 2006) and peninsulas (Rees et al. 2010) where continental margins are narrow. Alternatively, the inability to isolate why individual males migrated where they did may stem from not measuring the proper metric(s). Given that male foraging locations spanned the nesting range for this species in the NW Atlantic (NMFS and USFWS 2008), foraging ground selection may be genetically rooted (Bowen et al. 2004). However, genetic markers that are sensitive enough to assign individual turtles to nesting regions are not currently available. Because conservation concerns may not be evenly distributed along latitudinal or longitudinal gradients (for example, Ragland et al. (2011) reported elevated contaminant loads for northbound males in the present study), efforts to elucidate the factors responsible for habitat selection should continue to remain a high priority.
Similarities in post-breeding dispersal patterns by adult male loggerhead sea turtles in the present study and in Greece (Schofield et al. 2010a) suggest that seasonal migrations may be innate, as opposed to regionally developed, which has global conservation implications. At both study locations, synchronized dispersal of resident and migrant adult male loggerhead sea turtles occurred in May (Schofield et al. 2010a; Arendt et al. in press). At both study locations, males migrated <1,400 km and reached foraging grounds by the end of June. Migration direction was not significantly different in either study perhaps due to small sample sizes, but it is worth noting that migrants most often foraged at higher latitudes than where the breeding ground was located. Southwood and Avens (2010) suggest that seasonal migration of juveniles and adult females to coastal foraging grounds at higher latitudes in the northern hemisphere is related to increased nutrient availability in colder waters (Lalli and Parsons 1993). Dispersal direction bias to higher latitudes is also reported for male leatherback sea turtles (Dermochelys coriacea) migrating to pelagic foraging grounds in the NW Atlantic (James et al. 2005). In contrast, east–west movements are reported for male hawksbill sea turtles (Eretmochelys imbricata) in the Caribbean Sea (van Dam et al. 2008), where nutrient availability is similar across latitudes.
Foraging patterns differed remarkably between adult male loggerhead sea turtles in the present study and in Greece (Schofield et al. 2010a). In the present study, males remained at a single foraging ground until initiating winter movements, consistent with data for females in this region (Foley et al. 2008; Girard et al. 2009; Hawkes et al. 2011). In contrast, Schofield et al. (2010a) reported that males often occurred at multiple foraging grounds prior to as well as during winter. Males tracked by Schofield et al. (2010a) were predominantly distributed <1 km from shore in semi-enclosed water bodies where they foraged (90% fixed kernel density) over areas <60 km2. Only one male in the present study had a comparable foraging area (100% MCP) of 31 km2, which was also located within an estuary. Only one other male in the present study foraged in a comparable area (471 km2) to foraging areas (417–490 km2) calculated by Schofield et al. (2010a) for males in the Central Adriatic Sea. The largest foraging area (1,005 km2) calculated by Schofield et al. (2010a) was associated with a pelagic male during winter and was approximately twice as large as male foraging areas in the Adriatic Sea; in the present study we also noted that foraging areas doubled for two of three males tracked during winter. Nevertheless, the largest male foraging area calculated by Schofield et al. (2010a) was half the size of the largest foraging area in the present study.
Behavioral polymorphism in foraging areas used by adult male loggerhead sea turtles noted between the present study and in Greece (Schofield et al. 2010a) likely reflects the relative availability of neritic habitats <100 m deep. In the NW Atlantic and GOM, loggerhead sea turtles forage opportunistically, but generally consume demersal prey (Plotkin et al. 1993; Seney and Musick 2007). Demersal foraging is also reported for loggerhead sea turtles in the Adriatic Sea (Lazar et al. 2011), where most males foraged after migrating away from Greece (Schofield et al. 2010a). Although loggerhead sea turtle diets in the Mediterranean Basin include a significant contribution from pelagic items (Revelles et al. 2007; Casale et al. 2008), greater demersal foraging has been noted for larger turtles as foraging efficiency improves (Casale et al. 2008). As such, we suggest that when adult male loggerhead sea turtles occupy areas associated with vast tracts of neritic habitat (such as in the NW Atlantic, GOM, and Adriatic Sea), foraging occurs over more protracted areas than when neritic habitats are consolidated. The consistent dive durations for males located across a large latitudinal gradient in a diversity of habitats in the present study support this assertion, as do similar magnitude differences in foraging area sizes for adult females in the NW Atlantic relative to the Mediterranean Basin (Hawkes et al. 2011).
By identifying areas where male loggerhead sea turtles are likely to occur, at least through fall, our data provide a crucial backdrop for future experimental studies. For example, localized loggerhead sea turtle abundance has declined (Mansfield 2006), concurrent with a reported diet shift away from benthic invertebrates (Seney and Musick 2007), in at least one of two large estuaries (Chesapeake Bay, Delaware Bay) that historically supported large seasonal loggerhead aggregations (Spence 1981; Lutcavage and Musick 1985). However, in the present study, adult males predominantly occupied coastal waters outside of these estuaries, a pattern also noted for adult females (Hawkes et al. 2011) and observed in commercial fisheries (Murray 2011). As such, in the same manner that acoustic telemetry arrays (Fujioka et al. 2010) revealed a hydrographic basis for a perceived decline in southern bluefin tuna (Thunnus maccoyii) abundance, experimental satellite telemetry studies could help determine the extent to which changes in loggerhead sea turtle abundance in Chesapeake Bay are influenced by standardized sampling in historic areas concurrent with shifts in loggerhead distributions to outside of the sampling area. Akin to efforts to increase the duration of data collection by attaching satellite transmitters to post-nesting rather than inter-nesting females (Godley et al. 2008), attachment of satellite transmitters to male loggerhead sea turtles collected during non-breeding periods could also potentially result in greater temporal data coverage than reported here given that physical transmitter damage associated with breeding (Hays et al. 2007) should be minimized.
We thank R. Brill (Virginia Institute of Marine Science) and anonymous reviewers for critical edits to earlier versions of this manuscript and subsequent contribution to this manuscript. Funding was provided by NOAA Fisheries grant NA03NMF4720281, and this research was conducted under Section 10(a)(1)(A) permit #1540 and Florida Marine Turtle Permit #163. This is contribution #686 of the Marine Resources Division of the South Carolina Department of Natural Resources.